Joseph Rodriguez

Astrophysics @ UC San Diego

ML Engineering in Scientific Systems

End-to-end ML systems: data, training, deployment, maintenance

I build physics-informed ML systems for gravitational-wave science and space-adjacent engineering, linking population synthesis, domain adaptation, and cloud/HPC pipelines to ship reproducible results.

Key Achievements (Top Signals)

  • Physics-informed deep learning fusing COMPAS/COSMIC/POSYDON with LIGO/Virgo data to infer binary black hole formation channels.
  • AI-powered infra automation at General Atomics: compliance/reporting pipelines and network diagnostics with AKIPS/IPAM.
  • Data-driven ops at UCSD (SPACES): 2x application processing speed; 30% waste reduction; 890+ students/year on a $55K budget.

Experience

Undergraduate Researcher, UC San Diego (Apr 2025–Present)
Impact: Simulation-based inference pipeline aligning population synthesis with observed GW events; building domain adaptation and uncertainty-aware models.

  • Integrated COMPAS, COSMIC, POSYDON outputs as Bayesian priors for SBI
  • Domain adaptation to reconcile simulator vs. detector distributions
  • Epistemic/aleatoric uncertainty quantification and CE-efficiency analysis via cross-modal Transformer attention

Infrastructure Systems Intern, General Atomics (Jun 2025–Aug 2025)
Impact: Automated infra workflows and AI assistants, reducing compliance/audit overhead and accelerating diagnostics.

  • Provisioned and automated Windows/Linux VMs and AWS instances across lifecycle
  • Built PowerShell/Python/Orchestrator tooling to flag orphaned servers and auto-generate compliance reports (~70% audit-time reduction)
  • Implemented AI onboarding expert and real-time networking assistant integrated with AKIPS/IPAM
  • Led AI adoption demos for 30+ interns and panel for 150+ employees

Academic Success Program Coordinator, SPACES, UC San Diego (Jun 2024–Jun 2025)
Impact: Scaled UCSD’s largest equity-based textbook program with analytics and automation.

  • Served 890+ students/year; managed $55K budget and 1,700+ inventory items
  • Built Python/Excel/Kuali analytics; 2x application processing speed; reduced waste ~30%
  • Introduced equity metrics and data-driven purchasing; streamlined workflows via system migrations

Core ML Stack

Python · PyTorch · NumPy · SciPy
AWS (EC2, S3, Batch) · Docker · Linux · HPC
Data Pipelines · Model Evaluation · Automation

Technical Skills

  • Machine Learning: Physics-informed DL, simulation-based inference, Bayesian methods, Transformers, domain adaptation
  • Infrastructure & DevOps: AWS, Docker, HPC/SLURM, AWS Batch, Linux/Windows admin, IaC/automation (Python, PowerShell)
  • Data Foundations: SQL, data modeling, HDF5/Parquet, data visualization, pipeline optimization

Education

University of California, San Diego, B.S. Astronomy & Astrophysics (Expected 2027)

Mission & Outreach

First-generation student focused on making space science more inclusive. I want to advance autonomous space systems, mission-planning algorithms, and scientific discovery tools, while widening access to STEM.

  • Astronomy Club Newsletter Automation, View
  • JWST Gas Morphology Classifier, View
  • Renewable Energy ROI Analysis, View
  • OER Integration at UC San Diego, View

Connect